AI-Assisted Multiclass Kidney CT Pathology Classification for Urological Imaging: An Optimized Hybrid Deep Learning Framework

Öz

Kidney CT imaging plays a central role in the urological diagnosis of renal cysts, calculi, solid tumors, and normal renal parenchyma. Although deep learning has shown promise for automated CT-based kidney pathology classification, existing approaches frequently rely on single-architecture models and lack systematic hyperparameter optimization, limiting both performance and reproducibility. This study aimed to develop and internally evaluate an optimized hybrid deep learning framework for multiclass kidney CT pathology classification in urological imaging, focusing on cyst, stone, tumor, and normal categories. A curated multiclass kidney CT dataset of 9,574 images was split into training (6,368), validation (1,997), and test (1,209) sets across four classes: Cyst, Normal, Stone, and Tumor. Four convolutional backbones-ResNet50V2, EfficientNetB3, DenseNet121, and Xception-were fine-tuned under a two-stage transfer learning protocol. Hyperparameters for each backbone were independently optimized using the Optuna framework with AdamW optimization and label smoothing. Class imbalance was addressed using weighted loss. Test-time augmentation was applied at inference. Two hybrid decision strategies were evaluated: validation-driven weighted soft voting and a logistic regression stacking learner. This was an internal computational validation study using a same-source dataset split; no external cohort was available. On the internal same-source held-out test set, ResNet50V2, Xception, the weighted ensemble, and the stacking model achieved perfect performance (100% accuracy, precision, recall, F1-score, and one-vs-rest ROC-AUC). DenseNet121 and EfficientNetB3 showed near-ceiling results. These outcomes may reflect strong class separability within the analyzed dataset and should be interpreted with caution in the absence of external validation. Ensemble strategies provided consistent performance gains over weaker individual backbones, particularly for the Stone class. The proposed optimized hybrid deep learning framework achieved strong internal performance for multiclass kidney CT pathology classification across urologically relevant categories. The findings suggest that architecture diversity and systematic hyperparameter optimization can substantially improve predictive robustness. External multi-center validation is required before clinical implementation in urological imaging workflows. The framework may serve as a candidate decision-support tool for radiologist-assisted interpretation of kidney CT findings, provided prospective validation is conducted. A revision-stage integrity audit found no byte-identical, pixel-identical, or repeated Roboflow source-identifier overlap across the predefined splits; however, two cross-split pairs showed very high structural similarity (SSIM >= 0.995), and patient-level independence could not be verified because patient identifiers were unavailable. Accuracy uncertainty was additionally quantified using 95% Wilson score confidence intervals.

Anahtar Kelimeler

• Renal cyst
• Renal stone
• Renal tumor
• Optuna
• Test-time augmentation
• Clinical decision support

Etik Beyan

Ethics committee approval was not required for this study because it involved no animal or human subjects.

Teşekkür

The dataset used in this study was the public Roboflow Universe project "Renal- Failure-Analysis" maintained by Mini Project (workspace: mini-project-6ilyj; project: renal-failure-analysis-5ymli), version 4 (Mini Project, 2023). The working export contained 9,574 images in the Cyst, Normal, Stone, and Tumor classes and was publicly listed under a Public Domain license at the time of revision. The public project page does not document the original clinical institution, DICOM-level anonymization workflow, or patient/study identifiers. The authors analyzed only the rendered image export supplied by the repository and did not have access to original DICOM headers or identifiable patient metadata. Therefore, patient-level split independence and the original anonymization process could not be independently verified. ChatGPT was used solely for English translation, grammar correction, and linguistic editing. The artificial intelligence tool was not used to generate the research design, dataset, analyses, model outputs, tables, figures, or scientific conclusions. The authors reviewed and approved the final manuscript and take full responsibility for its content.

Kaynakça

1. Abd El-Aziz, A. A., Mahmood, M. A., & El-Ghany, S. A. (2025). EfficientNet-B3-based automated deep learning framework for multiclass endoscopic bladder tissue classification. Diagnostics, 15(19), Article 2515. https://doi.org/10.3390/diagnostics15192515
2. Aby, A. E., Salaji, S., Anilkumar, K. K., et al. (2025). Classification of acute myeloid leukemia by pre-trained deep neural networks: A comparison with different activation functions. Medical Engineering & Physics, 135, Article 104277. https://doi.org/10.1016/j.medengphy.2024.104277
3. Al-Adhaileh, M. H., Alsharbi, B. M., Aldhyani, T. H. H., et al. (2025). DLAAD-deep learning algorithms assisted diagnosis of chest disease using radiographic medical images. Frontiers in Medicine, 11, Article 1511389. https://doi.org/10.3389/fmed.2024.1511389
4. Al-Otaibi, S., Rehman, A., Mujahid, M., et al. (2024). Efficient-Gastro: Optimized EfficientNet model for the detection of gastrointestinal disorders using transfer learning and wireless capsule endoscopy images. PeerJ Computer Science, 10, e1902. https://doi.org/10.7717/peerj-cs.1902
5. Alzu'bi, D., Abdullah, M., Hmeidi, I., et al. (2022). Kidney tumor detection and classification based on deep learning approaches: A new dataset in CT scans. Journal of Healthcare Engineering, 2022, Article 3861161. https://doi.org/10.1155/2022/3861161
6. Aouadi, S., Torfeh, T., Bouhali, O., et al. (2024). Prediction of cervix cancer stage and grade from diffusion weighted imaging using EfficientNet. Biomedical Physics & Engineering Express, 10(4), Article 045042. https://doi.org/10.1088/2057-1976/ad5207
7. Badr, M., Elkasaby, A., Alrahmawy, M., & El-Metwally, S. (2025). A multi-model deep learning architecture for diagnosing multi-class skin diseases. Journal of Imaging Informatics in Medicine, 38(3), 1776–1795. https://doi.org/10.1007/s10278-024-01300-w
8. Boudi, A., He, J., & Abd El Kader, I. (2024). Enhancing Alzheimer's disease classification with transfer learning: Fine-tuning a pre-trained algorithm. Current Medical Imaging, 20, Article e15734056305633. https://doi.org/10.2174/0115734056305633240603061644

9. Chen, X., Pu, X., Chen, Z., et al. (2023). Application of EfficientNet-B0 and GRU-based deep learning on classifying the colposcopy diagnosis of precancerous cervical lesions. Cancer Medicine, 12(7), 8690–8699. https://doi.org/10.1002/cam4.5581
10. Chollet, F. (2017). Xception: Deep learning with depthwise separable convolutions. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 1800–1807. https://doi.org/10.1109/CVPR.2017.195
11. Esteva, A., Robicquet, A., Ramsundar, B., et al. (2019). A guide to deep learning in healthcare. Nature Medicine, 25(1), 24–29. https://doi.org/10.1038/s41591-018-0316-z
12. Gan, K., Liu, Y., Zhang, T., et al. (2024). Deep learning model for automatic identification and classification of distal radius fracture. Journal of Imaging Informatics in Medicine, 37(6), 2874–2882. https://doi.org/10.1007/s10278-024-01144-4
13. Han, S., Hwang, S. I., & Lee, H. J. (2019). The classification of renal cancer in 3-phase CT images using a deep learning method. Journal of Digital Imaging, 32(4), 638–643. https://doi.org/10.1007/s10278-019-00230-2
14. He, K., Zhang, X., Ren, S., & Sun, J. (2016). Identity mappings in deep residual networks. Computer Vision – ECCV 2016, 630–645. https://doi.org/10.1007/978-3-319-46493-0_38
15. Hossain, M. N., Bhuiyan, E., Miah, M. B. A., et al. (2025). Detection and classification of kidney disease from CT images: An automated deep learning approach. Technologies, 13(11), Article 508. https://doi.org/10.3390/technologies13110508
16. Huang, G., Liu, Z., van der Maaten, L., & Weinberger, K. Q. (2017). Densely connected convolutional networks. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 4700–4708. https://doi.org/10.1109/CVPR.2017.243
17. Kan, H.-C., Lin, P.-H., Shao, I.-H., et al. (2025). Using deep learning to differentiate among histology renal tumor types in computed tomography scans. BMC Medical Imaging, 25(1), Article 66. https://doi.org/10.1186/s12880-025-01606-3
18. Kreouzi, M., Theodorakis, N., Feretzakis, G., et al. (2025). Deep learning for melanoma detection: A deep learning approach to differentiating malignant melanoma from benign melanocytic nevi. Cancers, 17(1), Article 28. https://doi.org/10.3390/cancers17010028
19. Kumar, Y., Shrivastav, S., Garg, K., et al. (2024). Automating cancer diagnosis using advanced deep learning techniques for multi-cancer image classification. Scientific Reports, 14(1), Article 25006. https://doi.org/10.1038/s41598-024-75876-2
20. Mao, J., Du, Y., Xue, J., et al. (2024). Automated detection and classification of mandibular fractures on multislice spiral computed tomography using modified convolutional neural networks. Oral Surgery, Oral Medicine, Oral Pathology and Oral Radiology, 138(6), 803–812. https://doi.org/10.1016/j.oooo.2024.07.010
21. Meeran, S. B. U. (2025). Walrus optimization-enhanced ResNet-50 for AI-driven renal malignancy prediction with occlusion sensitivity-based interpretation. Asian Pacific Journal of Cancer Prevention, 26(8), 2995–3004. https://doi.org/10.31557/APJCP.2025.26.8.2995
22. Mini Project. (2023). Renal-failure-analysis dataset (Version 4) [Data set]. Roboflow Universe. https://universe.roboflow.com/mini-project-6ilyj/renal-failure-analysis-5ymli
23. Pimpalkar, A., Saini, D. K. J. B., Shelke, N., et al. (2025). Fine-tuned deep learning models for early detection and classification of kidney conditions in CT imaging. Scientific Reports, 15(1), Article 10741. https://doi.org/10.1038/s41598-025-94905-2
24. Rajinikanth, V., Vincent, P. M. D. R., Srinivasan, K., et al. (2023). A framework to distinguish healthy/cancer renal CT images using the fused deep features. Frontiers in Public Health, 11, Article 1109236. https://doi.org/10.3389/fpubh.2023.1109236
25. Rajpurkar, P., Chen, E., Banerjee, O., & Topol, E. J. (2022). AI in health and medicine. Nature Medicine, 28(1), 31–38. https://doi.org/10.1038/s41591-021-01614-0
26. Sathya, R., Mahesh, T. R., Bhatia Khan, S., et al. (2024). Employing Xception convolutional neural network through high-precision MRI analysis for brain tumor diagnosis. Frontiers in Medicine, 11, Article 1487713. https://doi.org/10.3389/fmed.2024.1487713
27. Shaheed, K., Abbas, Q., Hussain, A., & Qureshi, I. (2023). Optimized Xception learning model and XGBoost classifier for detection of multiclass chest disease from X-ray images. Diagnostics, 13(15), Article 2583. https://doi.org/10.3390/diagnostics13152583
28. Tan, M., & Le, Q. V. (2019). EfficientNet: Rethinking model scaling for convolutional neural networks. Proceedings of the 36th International Conference on Machine Learning, 97, 6105–6114. PMLR.
29. Topol, E. J. (2019). High-performance medicine: The convergence of human and artificial intelligence. Nature Medicine, 25(1), 44–56. https://doi.org/10.1038/s41591-018-0300-7
30. Wilson, E. B. (1927). Probable inference, the law of succession, and statistical inference. Journal of the American Statistical Association, 22(158), 209–212. https://doi.org/10.1080/01621459.1927.10502953
31. Xu, P., Zhao, J., Wan, M., et al. (2024). Classification of multi-feature fusion ultrasound images of breast tumor within category 4 using convolutional neural networks. Medical Physics, 51(6), 4243–4257. https://doi.org/10.1002/mp.16946
32. Yao, H., Tian, L., Liu, X., et al. (2023). Development and external validation of the multichannel deep learning model based on unenhanced CT for differentiating fat-poor angiomyolipoma from renal cell carcinoma: A two-center retrospective study. Journal of Cancer Research and Clinical Oncology, 149(17), 15827–15838. https://doi.org/10.1007/s00432-023-05339-0
33. Ye, D., Ou, Z., Ye, F., et al. (2025). Ensemble deep learning model for accurate assessment of renal fibrosis in chronic kidney disease using two-dimensional shear wave elastography images. BMC Medical Imaging, 25(1), Article 417. https://doi.org/10.1186/s12880-025-01964-y
34. Zhang, X., Lai, F., Chen, W., & Yu, C. (2024). An automatic glaucoma grading method based on attention mechanism and EfficientNetB3 network. PLOS ONE, 19(8), e0296229. https://doi.org/10.1371/journal.pone.0296229
35. Zhou, T., Ye, X., Lu, H., et al. (2022). Dense convolutional network and its application in medical image analysis. BioMed Research International, 2022, Article 2384830. https://doi.org/10.1155/2022/2384830